2. Literature Review

The literature review sheds light on the main theoretical and empirical contributions regarding the relationships among three major strands. We first focus on the theoretical foundation of the nexus between energy poverty and GVC participation; we then focus on their empirical relationships, and finally we examine the role played by the institutions in this nexus.

3. Data and Methodology

3.3 Model Specification

Based on the gap in the literature, this empirical study focuses on the role of GVCs and how the quality of institutions affects energy poverty at the global level, but specifically comparing both rural and urban areas.

Following previous theoretical and empirical contributions, we take advantage of the recent studies by Meka’a et al. (Reference Meka’a, Djamen and Noufelie2024) and Nkoa et al. (2024), who explained the dynamics and drivers of energy poverty in African countries, to build our empirical structure. We include the explanatory factors listed above of GVC participation and QI variables in order to ensure analytical rigor and comparability of international data. Our EP function takes the following form:

(1)\begin{equation}E{P_{it}} = f(GV{C_{it}};Q{I_{it}};{X_{it}})\end{equation}

where $E{P_{it}}$ measures the level of energy poverty of country i in year t, $GV{C_{it}}$ and $Q{I_{it}}{\text{ }}$capture the GVC participation degree of country i in year t and ${X_{it}}$ is the vector of baseline control variables. The econometric form of Equation (1) follows:

(2)\begin{equation}E{P_{it}} = {\beta _0} + {\beta _1}GV{C_{it}} + {\beta _2}Q{I_{it}} + \sum\limits_{i = 3}^6 {{\beta _i}{X_{it}}} + {\varepsilon _{it}}\end{equation}

where ${\beta _0}$ captures the unobserved parameters in our models, ${\beta _1}$ and ${\beta _2}$ are the parameters associated with GVC and QI respectively. ${\varepsilon _{it}}$ is the error term.

Equation (2) allows us to reach our first objective: to evaluate the direct influence of GVC participation on energy poverty. Furthermore, we include a multiplicative interaction term in equation (2) in order to reach the second objective that analyzes the mediating role played by QI in the GVC–EP nexus. Therefore, as suggested by Agradi (Reference Agradi2023), equation (3) includes the use of natural logarithms for all variables in order to eliminate the effects of data fluctuations and the potential heteroskedasticity. All other variables are converted to natural logarithms to ensure data non-volatility, with the exception of institutional quality. The final equation to be estimated is as follows:

(3)\begin{equation}\begin{gathered} E{P_{it}} = {\beta _0} + {\beta _1}GV{C_{it}} + {\beta _2}Q{I_{it}} + {\beta _3}{(GVC \times QI)_{i,t}} + {\beta _4}GD{P_{it}} + {\beta _5}Trad + \\ {\beta _6}FD{I_{it}} + {\beta _7}UR{B_{it}} + {\varepsilon _{i,t}} \\ \end{gathered} \end{equation}

where $L$ represents the natural logarithmic operator and $(GVC \times QI)$ is the interaction term between GVC participation and the quality of institutions.

The results obtained from the estimation of Equation (3) cannot draw definitive conclusions. As noted by Meka’a et al. (Reference Meka’a, Djamen and Noufelie2024), by deepening our investigation we clearly appreciate the results of the mediating factor. For this purpose, we determine the net (total) effect of the direct effect of GVC on energy poverty. Following Luo et al. (Reference Luo, Sun, Tao, Tan and Kamarudin2024), the net effect ( ${\text{GV}}{{\text{C}}_{NEF}}$) of GVCs effect on EP trough QI, is computed as follows:

(4)\begin{equation}GV{C_{NEF}} = {\beta _1} + {\beta _3}\overline {QI} \end{equation}

where $\overline {QI} {\text{ }}$is the mean of quality of institutions.

Finally, it is further relevant to compute the threshold effect of QI, which means the optimal QI value needed to change the sign of the GVC coefficient. The computation of QI thresholds is based on the following formula:

(5)\begin{equation}Q{I_{it}} = - \frac{{{\beta _1}}}{{{\beta _3}}}\end{equation}

However, two conditions are necessary to justify the obtained fit in the threshold calculation: first, at least one coefficient from the direct and indirect effects should be statistically significant; and, second, these coefficients should have opposite signs (Meka’a et al., Reference Meka’a, Djamen and Noufelie2024).

3.4 Empirical Strategy

This subsection provides more precision on our econometric path, which is divided into three steps: the first step is the cross-sectional dependence and slope heterogeneity test; the second evaluates the short- and long-run relationships; and, finally, a set of robustness approaches is provided to evaluate the consistency of the baseline results.

3.4.1 Cross Sectional Dependence and Slope Heterogeneity Test

One of the major problems during the estimation process of the panel data model is cross-sectional dependence (CSD), which may lead to biased results. Indeed, as African countries are similar in economic, cultural, and geographic aspects, it is also observed that the changes and disruptions observed in the GVCs in recent years have redefined the links among them (Agradi, Reference Agradi2023). Therefore, this study introduces Pesaran’s (Reference Pesaran2004) cross-sectional dependence test (because N (51) is larger than T (29). The formulation of this test is as follows:

\begin{equation*}{\delta _{CSD}} = \frac{{((T*N)(N - 1))}^{\frac{1}{2}}}{2}{\mathop {PR}\limits^{\hat{}}}_{N}\end{equation*}

with N, representing the cross sections, T the stated time, and ${\mathop {PR}\limits^{\hat{}}}_{N}$ illustrates the pairwise parameters.

Next, by considering the CSD issue, we test slope heterogeneity from Pesaran and Yamagata (Reference Pesaran and Yamagata2008), as follows.

\begin{equation*}\mathop {{\Delta _{SH}}}\limits^{\hat{}} = {(N)^{\frac{1}{2}}}*{(2k)^{\frac{1}{2}}}*(\frac{1}{N}\mathop S\limits^{\hat{}} - k)\end{equation*}

\begin{equation*}\mathop {{\Delta _{AdjSH}}}\limits^{\hat{}} = {(N)^{\frac{1}{2}}}*{\left( {\frac{{2k.(T - k - 1)}}{{T + 1}}} \right)^{^{ - \frac{1}{2}}}}*(\frac{1}{N}\mathop S\limits^{\hat{}} - k)\end{equation*}

The previous equations $\mathop {{\Delta _{SH}}}\limits^{\hat{}} $ and $\mathop {{\Delta _{AdjSH}}}\limits^{\hat{}} $, which take into account the delta and adjusted delta statistics, are made under the null hypothesis that ‘slope parameters are homogeneous across panel identities. $\hat{S}$, $k$, $T$, and $N$ respectively represent the Swamy’s statistics, regressors, time, and cross-sections.

3.4.2 Long-run Estimations

Given that a primary analysis has provided some relevant characteristics of our data, it is critical to choose the appropriate estimation method. Indeed, the occurrence of CSD has led to the abandonment of several recently adopted approaches to find energy poverty drivers. For instance, since our panel includes 51 countries over the 24 years (1995–2018), we cannot employ FGLS as did Meka’a et al. (Reference Meka’a, Djamen and Noufelie2024) and Nawaz and Rahman (2023), because the Park’s (1967) condition is not guaranteed (T large and N small). Additionally, although quantile regressions deal with heterogeneity, they fail to address the CSD issue. We follow Agradi (Reference Agradi2023), who uses DCCE-PMG to investigate how the long- and short-run remittances affect energy poverty in Africa. The author claimed that DCCE-PMG is more accurate than CCE-PMG because the dynamic specification addresses the endogeneity issue, as authors claimed a significant association in energy access between two consecutive periods. In addition, the GMM approach could solve the endogeneity issue, by including a dummy country that fails to address the cross-sectional issue (Hoechle, Reference Hoechle2007). DCCE is a substantial method to take into account both temporal and cross-sectional dependencies in a panel. Furthermore, DCCE-MG and DCCE-PMG estimators assume both heterogeneous coefficients in the short run; however, in the long run, the later assumes homogenous coefficients, while the former still assumes heterogeneous coefficients (Agradi, Reference Agradi2023). Indeed, short-run heterogeneity can be linked to diversity in institutions, economic development, culture, etc. However, all of these economies are likely to converge toward a steady state in the long run. Additionally, DCCE-PMG takes advantage of the Driscoll and Kraay (Reference Driscoll and Kraay1998) and DCCE-MG methods, as DCCE-PMG, by including cross-sectional dependence, symptomatically exhibits a differential effect due to unobserved common factors even if the number of unobserved common factors is unknown (Pesaran, Reference Pesaran2006). However, the Driscolland Kraay method is limited, especially in their assumptions, and it remains difficult to generalize, especially in the presence of unobserved common factors. Therefore, DCCE-PMG is a more efficient estimate to provide accurate results in the long run (Chudik and Pesaran, 2013). As Kim and Lin (2018) and Agradi (Reference Agradi2023) state, we agree that this approach appears to be the best choice for our sample as it accounts for the homogeneity across countries in the long run and countries specificities in the short run.

3.4.3 Robustness Check

To check the consistency and reliability of our findings, we introduce a set of robustness approaches that include the optional measures of our variables of interest, sub-sample robustness, and another estimation method.

5. Discussion of Findings

This section further discusses the findings and implications of the previous studies. The benchmark model shows that GVC participation has a negative effect on overall multidimensional energy poverty in Africa as well as in the short and long run. But there is a larger magnitude in the short run than in the long run. This means that integration into the GVCs will lead to a significant reduction in energy poverty in African countries, allowing us to validate hypothesis 1 of the current study. The theoretical mechanism underpinning this is as follows: participation in global production chains leads domestic firms to grow up in international competition in the energy development field, resulting in an investigation of the energy sector. Furthermore, the competitive and strategic advantages developed by Porter (1985) confirmed that natural resource-rich countries, such as African countries, are expected to improve energy access through extraction and export for the first time, and, in the long run, importing technology or refined products should mitigate energy poverty. Ongo Nkoa (2023) found a significant association between natural resources and energy poverty in sub-Saharan Africa. A further mechanism has been presented by Tinta et al. (Reference Tinta, Sarpong, Ouedraogo, Hassan, Mensah-Bonsu and Onumah2018) explaining that through countries’ integration, the access to space for ‘learning to compete’ and for ‘self-discovery’ offers new opportunities to local industries for greater rigor and competition in GVCs. Then, a share of the profit from GVC participation could be invested in the energy field to reinforce their position on the GVCs, resulting in energy poverty reduction. These findings are supported by previous empirical studies (Agostino et al., Reference Agostino, Giunta, Ruberto and Scalera2023; Obeng et al., Reference Obeng, Mwinlaaru and Ofori2021). Indeed, it has been shown that participation in GVC can reduce energy poverty. On the one hand, domestic firms gain from producing/using new renewable energy sources (Agostino et al., Reference Agostino, Giunta, Ruberto and Scalera2023). On the other hand, Delera et al. (Reference Delera, Pietrobelli, Calza and Lavopa2022) asserted that domestic firms would be forced to employ smart and digital technology to achieve the best efficiency when using clean energy sources, which are required on an international scale. Similarly, Ji et al. (Reference Ji, Liu, Wu, Su, Ye and Feng2022) and Luo et al. (Reference Luo, Sun, Tao, Tan and Kamarudin2024) found that participation in GVCs promotes renewable energy access through the opportunity to access new resources and organize production. Finally, Obasanjo et al. (Reference Obasanjo, Olayiwola, Okodua, Adediran and Lawal2021) and Zhang et al. (Reference Zhang, Lin, Wang and Shahbaz2024) associated it with the capacity to participate in GVCs that offer an easy way to join the global market using domestic firms, which is favorable for engaging the resources needed to curb energy poverty. Antras (Reference Antras2020) further argues that integration into GVC participation is critical to the emergence of energy technology in developing countries. In addition, because GVCs promote inclusive growth (Obeng et al., Reference Obeng, Mwinlaaru and Ofori2021), the spillover effect may generate more access to basic infrastructure, such as clean energy sources, in the long and short run. Furthermore, our evidence supports the pollution hallo hypothesis since GVCs are likely to reduce energy poverty in that strong environmental regulations will lead foreign firms and companies to use new technologies in all GVC participation segments, resulting in the promotion of clean energy value chains (Zhu et al., 2019).

Looking at the rural and urban heterogeneity outcomes, irrespective of the time period, the direct effect of GVC participation is still negatively linked to both urban and rural energy poverty, but with lower magnitudes in rural areas. This means that energy poverty reduction in rural areas is less associated with GVCs than in urban areas. This is plausible because rural space economies tend to be categorized by marginal production with inadequate government funding and a lack of modern infrastructure. However, we cannot find robust evidence regarding the explanation of rural–urban energy poverty inequality through GVC involvement. We further found that among the control variables, GDP reduces the rural–urban energy poverty gap, FDI and trade openness do not affect it, while the urban population impact is significant but only in the long run. These findings were supported by those of Meka’a et al. (Reference Meka’a, Djamen and Noufelie2024).

Additionally, the findings reveal mixed influences of QI in the GVC-MEPI nexus: significant and negative for overall MEPI, significant and positive for urban areas, and negative and non-significant for rural areas in the long and short run. On the one hand, GVCs play a significant role in the overall reduction of multidimensional energy poverty in African economies through better QI. Therefore, enhancing the quality of institutions is critical in supporting overall clean energy development through the participation in GVCs in Africa. On the other hand, the mediating role of QI remains low in significantly reducing rural energy poverty. In contrast, QI leads to reinforcing urban energy poverty related to an increase in GVC participation, and this adverse result has been found as a consequence of rapid urbanization, which can mitigate institutional effectiveness and increase energy poverty (Ongo Nkoa et al., Reference Ongo Nkoa, Tadadjeu and Njangang2023). This evidence is consistent with that of Agostino et al. (Reference Agostino, Giunta, Ruberto and Scalera2023), who showed a significantly positive effect of GVC participation on energy development. When countries feature relatively better institutions, they should support domestic firms’ citizenship value chain segments, resulting in a reduction in their lack of access to energy in the long run.

Recently, Ghosh and Dutta (Reference Ghosh and Dutta2022) confirmed similar evidence and recommended that developing countries should implement a combination of environmental standards, strong regulatory conformity, and virtuous energy management policies to achieve zero-energy post-targeting. We further showed a positive association between QI and GVCs, as in Barbero and Rodriguez-Crespo (Reference Barbero and Rodriguez-Crespo2020). Africa’s GVC involvement presents great potential for the development of renewable energy, such as biomass, solar, wind, and hydropower, which represent attractive factors for investors in renewable energy development (Aliyu et al., 2018). This is relevant because the development of green energy sources can limit harmful emissions and provide sufficient energy for the population to preserve the environment. Second, the state of democracy in these countries is relatively low in capturing the net effect of GVC on energy access.

Similarly, the findings confirm that through QI, GVC participation leads to a reduction in the rural–urban gap in energy poverty. Checking for robustness, the MMQR approach supports this evidence and indicates an asymmetric effect of a larger magnitude for countries with a low MEPI. That is, a non-linear relationship occurs between GVC participation and MEPI and between GVC. The results are also robust regarding alternative measures of GVC participation, including FVA or Domestic value-added, and QI proxies. Therefore, engaging in forward participation within GVCs – access to the information required about technology from trading partners to trade intermediate goods – may be a great avenue for energy development. In other words, African countries’ exports are related more to natural resources than to agricultural and local products/services. According to Obeng et al. (Reference Obeng, Mwinlaaru and Ofori2021), this may be linked to the fact that Africa’s economy features relatively few downstream firms that could easily and steadily transform imports from foreign economies into semi-finished and finished goods. In contrast, the DVX impact prevails over a long period when the mediating channel of QI is included. That is, the above transformation process occurs if better institutions exist. Reforms and adequate institutions specific to the agricultural sector should be included in countries expected to promote both GVC participation and clean energy development.

Although most public funding is directed toward major electricity grid extension and support projects, the effects are not observed in the long run. Lack of government support could increase energy poverty related to GVCs in both urban and rural areas and increase the rural–urban gap in energy poverty. Therefore, in countries with better quality of institutions, especially less corruption and better quality of regulation and governance, increased participation in GVCs will have a significant influence in facilitating access to electricity, clean fuel, and technology for cooking. It has been observed that promoting good institutions is an avenue for enhancing the diffusion of information and new technology (Nunn and Trefler, Reference Nunn and Trefler2014). Thus, institutional improvement, which could induce a better investment climate, increase private and foreign investments, and increase energy use, is expected to reduce energy poverty (Akalin and Erdogan, 2020).

Therefore, policymakers are confronted with the challenge of properly regulating and strengthening GVCs linkages’ to promote technology transfer and a better institutional environment while simultaneously ensuring clean energy access in a country. In addition, many scholars advocate more potent trade, domestic industrial, and innovation policies to drive energy upgrades. GVCs scholars, including Pietrobelli et al. (2022), have attracted fine-gained microeconomic focus. The reduction in Africa’s MEPI through GVC and the quality of institutions will obviously pass through combining upstream rather than downstream positions on value chain participation and a better quality of regulation and governance efficiency.

6. Conclusion and Policy Implications

Accessibility to basic services, such as energy, remains obvious despite the significant progress recorded in recent decades. Several empirical contributions have been followed to provide governments with some policies regarding the energy factor; however, there is no consensus, particularly in Africa, where energy poverty is still a significant problem. In this vein, the current study assesses the impact of participation in the GVC and quality of institutions (QI) on multidimensional energy poverty (MEPI) in Africa. For this purpose, a panel sample of 51 African countries over the period ranging from 1995 to 2018, and the Dynamic Common Correlated Model (DCEE-PMG) estimation method were used. The measurement of total GVCs includes both forward GVC participation and backward GVC participation; QI is the index from policy V and MEPI, which includes the share of the population without access to electricity and without access to clean fuel and technology for cooking (CFT). Findings show that GVC participation negatively affects energy poverty not only in the short run but also in the long run, meaning that GVCs reduce multidimensional energy poverty in Africa.

The effect of QI on the MEPI did not appear to be significant. However, regarding the role of QI in the GVCs-MEPI nexus, the results suggest that the reduction in energy poverty related to participation in GVCs would be reinforced through the institutional quality of African countries. That is, the better the quality of institutions, the larger the total effect of GVC participation on the reduction in energy poverty. Furthermore, QI is significant in GVC rural energy poverty, while it turns positive in urban areas. Additionally, the findings confirm that through QI, GVC participation leads to a reduction in the rural–urban gap in energy poverty. Checking for robustness, the MMQR approach supports this evidence and indicates an asymmetric effect with a larger magnitude for countries with a low MEPI. The results are further robust regarding alternative measures for GVC participation, including the FVA, DVX, and QI proxies. In light of our findings, evidence suggests an urgent need to reinforce the application and promotion of better institutions, essentially in clean energy sectors, to support the curb of energy poverty in both the short and long run. Recognizing the crucial contribution in defining the GVC participation terms and conditions led by firms and governments, the promotion of the best institution is likely to attract partnerships in the energy field and upgrade the local supply chain. Urban heterogeneity should be considered in government policies. Government policies related to stimulating GVC participation should be combined with institutional quality to increase investment in both electricity and clean fuel and technology for cooking and the sustainable transition toward clean and reliable energy.

Furthermore, we found that FVA led to reduced MEPI irrespective of the time period, whereas it occurred only in the long run under DVX. Quality of regulation and government efficiency are likely to channel GVC participation to decrease energy poverty. This is because African countries’ GVC participating firms are largely downstream and agriculture-based, particularly in rural areas, and thus take place in domestic value additions, resulting in exports to upstream industries in developed countries. Moving on from the above, policymakers should harmonize GVC participation and QI in order to reduce energy poverty. Indeed, our analysis validates the effectiveness of participation in GVC in the reduction of energy poverty in African countries. Government policies related to stimulating GVC participation should be included as a political ingredient for the strengthening of both imports of intermediate goods, their conversion into semi-finished and finished goods, and these last will finally be exported to foreign countries. Policies aimed at encouraging participation in the GVCs require particular institutional reforms according to the area in order to significantly reduce the rural–urban gap in energy poverty.